I Was Wrong About ARM Holdings

All the Intel (Nasdaq: INTC  ) bulls are going to love this: I was wrong about ARM Holdings (Nasdaq: ARMH  ) .

I now realize that I've been overly bullish on the British chip designer, as I had bought shares personally back in September. My thesis was centered on ARM's unique role in the mobile-chip supply chain as the dominant IP licensor of nearly all mobile processors found in smartphones and tablets today.

On top of that, the company has promising prospects in other markets like smart TVs, servers, and embedded devices.

I've since sold my shares, and here's why.

Tale of the tape
Unfortunately, there is a glaring disconnect between ARM's royalties and the mobile-device market that I can no longer ignore. In fact, it's this same discrepancy that led to an official sell recommendation in Stock Advisor. Our flagship service had recommended buying ARM as early as 2003 near $5, and it was subsequently crossed off the scorecard in June 2009.

It's true that ARM has risen substantially since then and would have greatly benefited the SA scorecard if it had kept its spot, as shares have gained over 350% since that sell recommendation, but that doesn't address the flaws in its royalty model. This is how shares have performed since getting the SA boot.

ARMH Chart

 

ARMH data by YCharts

The problem then -- and now -- is that ARM has only modestly benefited from the meteoric rise of mobile-device adoption, and shares carry a rather lofty valuation that's increasingly hard to justify, even as it gains traction in new markets.

These numbers don't add up
ARM realizes revenue on chip shipments one quarter after they're shipped. So, for example, chips shipped in the second quarter of any given year will show up in ARM's results in the third quarter.

In ARM's first-quarter report, it saw 1.1 billion chips shipped bound for mobile devices, including phones and tablets, which was actually a decrease compared to the 1.15 billion mobile-bound ARM chips a year ago. Remember, those correspond to fourth-quarter 2011 device shipments.

In contrast, Gartner's fourth-quarter figures show that worldwide mobile-phone sales grew more than 5% to 476.5 million. Within those estimates, smartphone sales soared 47% to 149 million. IDC's tablet figures show even higher growth, jumping 155% to 28.2 million units.

So the broader mobile-device market is growing, driven by incredible gains in smartphone and tablet adoption, yet the number of royalty-bearing ARM chips decreased over the same time.

ARM inside
Smartphones and tablets predominantly carry ARM chips. The first Intel Atom-powered smartphone just launched in the Indian market, and Intel currently has no tablet spots. MIPS Technologies (Nasdaq: MIPS  ) has a tiny sliver of the market, notably in low-end devices that are ultimately doomed. MIPS has a handful of smartphone wins, but don't expect to be familiar with any of them.

All the gadgets you know and love, notably Apple's (Nasdaq: AAPL  ) iPhone and iPad, carry ARM-based chips -- anything from NVIDIA Tegras to Qualcomm (Nasdaq: QCOM  ) Snapdragons to Texas Instruments OMAPs, among many others. So why isn't ARM seeing more upside as its partners pump out processors based on its designs?

Qualcomm's chip business was literally responsible for all of its growth last quarter. Its CDMA technologies segment, which includes ARM-based Snapdragons, jumped 56%, while its licensing segment was flat.

Apple continues to evolve its own ARM-based chip strategy using its A-family of chips in iPod touches and Apple TVs, alongside iPhones and iPads. Good luck getting any detail on ARM and Apple's relationship, though, for Apple's stringent supplier secrecy requirements also apply to ARM.

At Mobile World Congress last month, when asked about the pair's partnership, an ARM strategist simply replied, "We don't comment on Apple." When asked a follow-up question, he responded with British charm, "We don't comment on Apple. Full stop." (That's a British colloquialism for "period," as in, "This conversation is over. Period.")

The best way to play ARM is not ARM
One theory might be that ARM smartphone chips are simply cannibalizing ARM feature-phone (dumb phone) chips, so ARM's results may be more closely tied to the broader market, rather than smartphones and tablets specifically. (The smartphone and tablet figures above add up to only 177.2 million, compared to the 1.1 billion chips ARM reported.)

Still, ARM should have more to show for it, as smartphones carry more ARM chips than feature phones, and tablets are a nascent market that is more of a threat to Intel's PC business than anything else.

Seems like the processor IP licensing business may not be all that attractive, as MIPS doesn't look good, either. This is where the Intel bulls predict that vertically integrated Chipzilla will overtake the mobile market, but that's still a steep hill to climb. It can be done, but I still think the odds are stacked heavily against Intel in mobile.

ARM chips still own the market, but it just seems that ARM isn't the one enjoying most of the benefits from that domination. There are many rungs of the supply chain where companies have a chance to add value, and it looks like most of the value-adding takes place after ARM's role and is being enjoyed primarily by its partners. In fact, licensees like Qualcomm and Apple are doing just fine.

Another one of ARM's partners has significant upside potential as it begins to gain traction in "The Next Trillion-Dollar Revolution." This company continues to score processor design wins and will be cashing in thanks to ARM's designs. Get the free report now.

Fool contributor Evan Niu owns shares of Apple, but he holds no other position in any company mentioned. Click here to see his holdings and a short bio. The Motley Fool owns shares of Apple, Qualcomm, and Intel. Motley Fool newsletter services have recommended buying shares of Apple, Intel, and NVIDIA. Motley Fool newsletter services have recommended writing puts on NVIDIA. Motley Fool newsletter services have recommended creating a bull call spread position in Apple. The Motley Fool has a disclosure policy.

We Fools may not all hold the same opinions, but we all believe that considering a diverse range of insights makes us better investors. Try any of our Foolish newsletter services free for 30 days.


Read/Post Comments (11) | Recommend This Article (8)

Comments from our Foolish Readers

Help us keep this a respectfully Foolish area! This is a place for our readers to discuss, debate, and learn more about the Foolish investing topic you read about above. Help us keep it clean and safe. If you believe a comment is abusive or otherwise violates our Fool's Rules, please report it via the Report this Comment Report this Comment icon found on every comment.

  • Report this Comment On April 25, 2012, at 12:03 PM, SSchlesinger wrote:

    Good article. There is a difference between saying that ARM isn't a good company and saying that ARM is a good company but it's not about to go into hyper growth. The stock is sold on the story that it's about to conquer Intel in PC and Servers and soon ARM will be the only tech processor company. It's a fairy tale. I like my fairy tales to come from Disney.

    I believe a fair valuation of the ARMH stock is around $6. Everything else is a premium for profits you realistically won't see. ARMH is doing fine, it's a good company that makes nice products, but it needs a forward P/E closer to the rest of the industry of between 12 to 15.

    As for Intel, I can understand your skepticism about their position in mobile. Yes, they do have an uphill battle. But if you look at the history of Intel you'd find that uphill battles are the things they excel at. Stagnation is what kills Intel. When they have a target such as replacing ARM in the top spot of mobile processors for phone and tablets then they get very focused and very driven. A year ago people would have laughed if you mildly suggested Intel has a shot in mobile. Now, they aren't laughing anymore.

    The problem with Apple right now is that they know they are backing the wrong team. ARM was a good fit, initially for Apple. Intel didn't have the solution at that time to take Apple where it wanted to go with the iPhone and iPad. Apple went with the best processor at the time for these applications, which was ARM. Had Steve Jobs not picked ARM at that time I don't know if the smart phone and tablet markets would have been what they are today.

    But,from this point going forward there are HUGE problems with ARM that Apple can't ignore. The foundries like TSMC have made it clear that they don't wish to be exposed to risk. They are pushing their risk exposure onto the licensees such as Nvidia and Qualcom. TSMC has had problems shipping 28nm processors in significant quantities. It's very much a fabrication crises that has become more clear over the past few weeks.

    Intel has this Tic Tock strategy which is essentially a race to thinner and thinner nodes at the foundries. It appears that TSMC, GF, and Samsung don't feel the sense of urgency quite like they should. If Apple sticks with them they can face a reality where Intel based smart phones running Android will simply spank them in every test possible. When it becomes painfully obvious to non-tech oriented consumers that Apple doesn't have the best technology i the market they can lose market share. Seeing as Tim Cook is more of a logistics guy than a visionary this probably won't happen.

    Apple was pushing the PowerPC processor for years, another RISC processor on their computer side. They would tell their fanbase that these slower processors were really twice as fast as the numbers suggested. So a 1.5Ghz processor was really faster than an Intel 2.5ghz Pentium D. Turns out that the performance bump was more like 15% for risk not 100%. So even with the efficiencies the 1.5ghz processor was remarkably slower. Apple finally gave up on RISC and went with Intel. And, most likely history will repeat, eventually.

  • Report this Comment On April 25, 2012, at 12:36 PM, Foolme2x wrote:

    What would be a "followup question" to "no comment"? Just how incredibly thick are you? And given that, why would anyone be interested in your views on anything?

    In the spirit of full disclosure, I was long ARMH well before it was recommended by SA in 2003, and still am. Compound annual price growth has been 22.8% since inception. Throw in the dividend & total annual return has been 23.5%.

    Yes, it's been "dead money" for the past two years. Pundits like Jim Cramer pumped it as an "Apple derivative" play, and the stock price overheated in 2010. Surface skimmers like Cramer never looked deeply enough into anything to really understand the companies and businesses that they talk about. As is correctly noted above (but with incorrect interpretation), Apple is an important but still relatively minor (~10%) part of ARMH's business. If Apple continues to boom, ARMH will continue to benefit.

    And Apple will continue to soar as long as the global consumer associates status with having paid an ultra premium price for anything with an Apple logo on it. Logic would seem to dictate that Apple's run must eventually end. But I wouldn't short Apple. To paraphrase, the consuming public certainly has the capability to stay irrational for at least a day longer than I could stay solvent if I bet against Apple.

    On the other hand, if/when Apple eventually falls from favor, ARMH will do just fine. Partly because it will most likely be there in whatever product the consumer chooses over Apple, but more importantly because the other 90% of its business will continue to grow.

  • Report this Comment On April 25, 2012, at 1:18 PM, Foolme2x wrote:

    SSchlesinger - Intel has "owned" the PC market for 30 years and the server market for more than a decade. Outside those two giant areas, they've always been an also ran. They've been the 800 lb gorilla in the corner of ARM's markets for years. They have been specifically chasing ARM in the low power consumption market for the past 5 years. Wake me if they ever get closer than a year behind.

    BTW, both INTC and ARMH have been in my portfolio for more than 10 years. All things considered, INTC has done OK, giving CAG of about 8% including dividends. But the CAG of the ARMH investment has been nearly 3 times that.

  • Report this Comment On April 25, 2012, at 2:41 PM, SSchlesinger wrote:

    Look up Nvidia 680 on Newegg and you'll see more than a month after it's introduction that except for one model, it's vaporware. So what, who cares about a $500 video card? It has the TSMC's 28nm GPU in it. What's this market? 20K units worldwide? And they still can't supply this?

    That is the problem in a nutshell. Sure you'll find lots of interesting ARM processors with great specs and great performance benchmarks. You just won't be able to buy them in significant quantities. ARM is going to lose their top spot to Intel on a manufacturing basis. While ARM will have the best specs in the market for sometime to come, Intel will have the best SoCs available.

    ARMH has taken its battle to the media, while Intel is taking it to the streets. I wonder who will win. This is a good time to dump your ARMH stock as the author suggested. Or, you can wait until the institutional investors drive the stock down to an industry standard 12 to 15 forward P/E making it a decent $6 stock.

  • Report this Comment On April 25, 2012, at 5:56 PM, ibankingcrooks wrote:

    @SSchlesinger: The video chip you are discussing - a Nvidia 680 - is not an ARM product. It has nothing to do with ARM at all. You need to learn a little bit about processor architecture before you go worrying about the supply chain around them.

    Furthermore ARM has *nothing* to do with process (something everyone likes to talk about since it involves simple numbers like 28 and 40. "Lower is better"). As you seem aware - process involves fabs: But Intel's competition is TSMC and GlobalFoundries, the largest fabs in the world. Furthermore, process matters less and less as you go higher up the application stack. What??...

    To translate: its not important in real-world terms. Analysts *love* to talk about process. But it doesn't matter. Don't believe me? What is the feature size on the Apple A5X chip? 32nm. Where is it made? Samsung in Austin. Wait, wait... what about Intel. Intel's factory just doesn't matter because they don't have the technology to deliver that performance at those power levels. So who care what their process is? Well, only analysts do.

    -----

    To value ARM correctly you can not look at the revenues this year or next - and then do a pretty CAPM on that to decide where the stock should be, unless you understand the sales process.

    ARM's best designs are being sold right now. Revenue will not be collected until 2-3 years from now. Who knows what the process will be then. (And who cares) We can be sure of one thing: Intel will have the best process, TSMC second best, and GlobalFoundry third. With Samsung and a few others having much smaller facilities with process near Intel's.

    Anyway... that's not the point. The point is ARM's best design's (multi-core, Mali, 64-bit, ARM 8 architecture) will not see real revenue until 2015-16. And process makes *no difference* to them. They are going to get revenue b/c chip makers have *already committed* to their design.

    I suggest you go here to understand how ARM actually makes its money:

    http://en.wikipedia.org/wiki/List_of_ARM_microprocessor_core...

  • Report this Comment On April 25, 2012, at 6:22 PM, ibankingcrooks wrote:

    @SSchlesinger: I am sorry - I didn't see your earlier comment.

    It appears you *think* you understand processor design.

    Here is why Intel won in the RISC/CISC games in the mid-90s (btw, most processor's today can't really be categorized like this anymore)

    RISC versus CISC, from a manufacturing standpoint - which appears to be your main concern - is about die size. From a software standpoint it means more instructions which do more per cycle = equal faster CPU. That's CISC. CISC also requires much large die size to make hardware with all those cool instructions.

    When the die is bigger, you need more power. In the 90s, people bought desktops. So Intel's solution was two fold: add more instructions that do more per cycle (MMX, SSE, SSE2, blah blah) and race to a newer process fast so that die size could be decreased, and speed increased. It was a fantastic strategy. It worked awesome. And guess what: there was a POWER CABLE.

    Laptop came... people seemed ok with having to plug their laptop in now and then as long as the latest version of Windows or OS X was not too slow... years go by... But now, enter the world of mobile. People don't want to plug in their phone every 5 hours. They don't think that's acceptable. Also, they don't need *massive* amounts of processing power. What they need is good processing balanced with excellent power characteristics.

    In this world, die size is still important, and so is feature size. But for different reasons. What's most important is an efficient CPU. A die size that's larger, but its 50% off (zero power) when it's not needed is much better than a die size which is a bit smaller, faster, but burns more power all the time (Intel Atom and friends).

    There is also a fundamental problem beyond this for Intel, and this is 'design wins.' When you pick a processor to center an embedded device around, you look at a lot of things. (I am a guy who would picks the processor - this is what I do, design devices). About the last thing we look at is 'feature size/process' or it's SPECint score. In fact, I doubt most engineers on a smart phone can even tell you about the process used in the fab for the CPU. They would have no idea.

    Instead we look for things like built-in radios, specialized accelerators, lots of IO, USB this and that, power management units, and all kinds of things to decrease the amount of soldering and individual parts on the board. Decreasing size and cost. And we look at chip cost. Now, I can choose from 20+ companies. OR I can choose Intel. Intel makes a few parts, 20+ companies make lots of parts, lots of choices, some close to exactly the specific parts we need.... On top of that everything Intel makes is expensive. See the dilemma? We pick a CPU based on cost, specific needs, power consumption, and then - finally - speed.

    Intel's entire strategy - it's entire business is setup for a world of:

    - Highly standardized architectures (where every device is the same).

    - Lot's of electrical power.

    - Sacrifice power for speed.

    - Dedicated CPUs, not SoC (system on chip).

    But this is exactly opposite of the mobile world. They are trying to change. But it's going to be very painful for them.... They also have lot's of capex, which ARM doesn't have. So you see this battle is not about ARM versus Intel. It's about the industry, and the marketplace, versus Intel. Intel has a serious up-hill battle in mobile.

  • Report this Comment On April 26, 2012, at 2:03 AM, SSchlesinger wrote:

    So, are you saying that Medfield doesn't fit in the power envelope that Intel claims? You've seen the Lava phone?

    http://www.anandtech.com/show/5770/lava-xolo-x900-review-the...

    I know you are a designer. Intel has a lot on staff as well. They designed a reference smart phone and are giving away the design for free. A mobile carrier can build these phones for under $200 each. Sure a quality ARM phone is only a few hundred dollars more and maybe will have a bit more impressive specs. But the carriers are getting squeezed. For under $200 you can't beat it. They are very good at pushing phones on people, especially ones that will help with their margins. This will be the free smart phone they give away with a 2 year contract. The iPhone and premium Android phones will continue to be the ones you have to pay for out of pocket AND sign a two year contract. From what I can see, once consumers get their hands on this many of them are going to be hard pressed to fork over money for something slightly better.

    Oh, and Intel will help pay for advertising. It's something that ARMH can't and won't do.

    But don't worry, ARMH has Windows to look forward to this year. They will break the Wintel stranglehold and be in 20% of all Windows based PC's and then move into servers for a full body slam? Right?

    http://www.extremetech.com/computing/127513-the-premature-de...

    Thanks for your comments, I don't understand processor design, but i do know enough to think that just maybe your view of Intel is a bit outdated.

    I am very clear on one thing. Intel goal isn't to put ARMH out of business. Their goal is to minimize the threat of ARM getting into their core business like they've been talking about for the past few years. Right now Intel is taking the battle to the smartphone and tablet instead of waiting for ARM to bring it to the PC and server markets.

    Right now in the mobile industry the threat of Intel is setting the pace. The problems with TSMC at 28nm wouldn't look so bad if their customers weren't screaming for product. Why? Paranoia about Intel perhaps? Intel has an uphill battle in mobile? It certainly doesn't look that way to me. It looks like the threat of Intel in mobile has already had a profound impact on the industry.

  • Report this Comment On April 26, 2012, at 2:06 AM, nivaga wrote:

    Firstly, in the interests of disclosure, I am long on ARM. ... and currently smarting from the 'temporary' drop :-)

    While there is some merit to the arguments of this article and SSchlesinger points, but, also some items that need to be taken into account:

    1. Most focusses on Intel's latest and future, vs ARMs existing. ARM is not standing still either!

    2. Fabrication is not the reason for ARM's current market lead in mobile over Intel - it is DESIGN. The majority of Intel's current war drum beating is around fabrication process. Fabrication is not a significant barrier to entry. That is a brute force approach.

    3. The lead time in mobile devices coming to market is 2-3 years. The existing mobile eco-system is invested in ARM and in some cases competes with Intel head on (e.g. Samsung). Intel are having to sell a place on a roadmap that already takes into account mutiple memebrs of the eco-system. I think the evidence of this is Intel's choice of non major player that it could boss and buy its way into the market with - and 2nd choice Lenovo for the next phone.

    4. I think the Lava Xolo smells like a glorified PR exercise in a market where it is unlikely to be found out etc. I hate to say this, but the AnandTech review also sounded like a massive shill often going out of the way to 'Normalise' things for comparison to make it look better. I have yet to see any review comment on another smartphone getting so hot that it gets throttled even 'only 1x' during a review. SSchlesinger, I think your comment "ARMH has taken its battle to the media" is more applicable to Intel currently!

  • Report this Comment On April 26, 2012, at 5:38 AM, winklerf wrote:

    @ibankingcrooks

    The lesson of the tale of the nVidia 680 is 28nm TSMC. It is an awful technology that doesn't yield and is going to weigh down TI, Qualcomm and nVidia. Their newer processors are designed for that process and will not be available to compete against Medfield, certainly not at a price that many will be willing to pay. By the time they get it straightened out, they will be competing with Intel's 22nm Silvermont.

    Now, consider that Medfield out CPUs everything but the Tegra 3, a TSMC 28nm chip. The Medfield core is really just a shrunk version of the original Atom core on a 32nm SoC. What is going to happen when the re-architected 22nm core shows up attached to Intel's rapidly improving graphics cores?

    Process technology is the whole game in the microprocessor business when you're shipping billions of units. Intel beat their competition in PCs and Servers by yielding a lot of die per wafer and either being a lot less expensive (in the case of servers), or able to pack in more features and lower power (in the case of desktops and notebooks). It is going to be very difficult for the ARM manufacturers to make a profit while they are either selling a larger die or fewer features. Their margins are going to start looking like AMD's, which they definitely aren't used to.

    Intel is most certainly in the position of having to scale down their processors and having to change the way they look at power. However, they have been doing this for quite a while. Their biggest problems with the original Atom were not so much with the processor core, but with the accompanying chipset, which was not integrated and was being manufactured on older processes.

    Further, the ARM designers are not as well prepared to scale up their processor designs. Having a quad core cell phone processor is ridiculous. You need four threads to take advantage of your processing power. Four cores are seldom used on a desktop, much less a cell phone or tablet. The only reason why this is happening is that ARM doesn't have meaningful SIMD instructions nor a robust enough data path to feed such an animal. Therefore, they've taken the less effective approach. They aren't even doing anything meaningful with regards to fine control of the processor clock. In short, they have a long way to go in terms of improving future performance. Going to 8 or 16 cores just isn't going to work out for them when 4 is already too many.

  • Report this Comment On April 26, 2012, at 6:20 AM, winklerf wrote:

    @nivaga

    ARM has been ahead in mobile design, because the smart phone has been a natural evolution from the base expertise of the ARM core (power) and the ARM designers (integration). That is how things have stood in the past. However, the characteristics of the processors in smart phones are growing into Intel's domain. They have had to make adjustments with respect to power and integration, but ARM will have to make adjustments with respect to performance in the future that Intel has already mastered and has a prohibitive patent portfolio in.

    It is fair to say that Intel still needs work in the power and integration departments, but it is my contention that they are much closer to ARM in those areas than ARM is to Intel in performance. To wit, Intel will have an extremely mobile processor (cell phone or smaller power draw) with the capability of 2010+ PC before ARM will, and will continue to have enormous advantage going forward from there. People are going to want that device as more AI gets integrated into their mobile computing device.

    What the history means with respect to process versus design is that ARM being ahead in design is an illusion. The future of mobile computing will be just as dependent upon performance as power efficiency and integration. The fact that Intel didn't get down the power efficiency and integration curves into cell phone land until now says as little about the future as ARM's lack of a 64 bit processor.

    As for ecosystems, the app ecosystem for cell phones and tablets isn't a meaningful barrier to entry. We're talking about mostly redundant, easy to develop apps that are mostly written for an interpreter. There is a reason Google and Amazon were able to jump straight in. There is very little to it. Most of the value of a smart phone right now is in the browser. The serious stuff hasn't even been written yet because it would overwhelm the phone. There is very little software development time invested in ARM architecture, most is platform independent, and even more is useless crap. The barrier to entry is nothing like Windows and Linux x86 where many feature rich applications exist that will not just port themselves over to a new platform.

    As for Lava, they are just the first to market. There will be similar releases by Lenovo in China and Orange in Europe. Intel is going harder after less developed markets like India and China because they don't have built in bias and are potentially very large. Intel can just compete on providing a quality product at a good price.

  • Report this Comment On April 26, 2012, at 10:14 AM, SSchlesinger wrote:

    @ fidgewinkle

    "I hate to say this, but the AnandTech review also sounded like a massive shill often going out of the way to 'Normalise' things for comparison to make it look better."

    "Firstly, in the interests of disclosure, I am long on ARM."

    Yeah, we can tell. BTW, AnandTech has a lot of credibility in the tech industry. But if you insist they are paid shills to prove your argument then so be it.

    " I think the evidence of this is Intel's choice of non major player that it could boss and buy its way into the market with - and 2nd choice Lenovo for the next phone."

    You're right, and guess what the next stop will be with their $200 smartphones? That's right, the cell phone carriers themselves. It kinda sucks when you have a competitor who doesn't play by your rules. Oh, and a another little known company signed a multi year multi phone deal, Motorola, you've maybe heard of them?

    " Fabrication is not the reason for ARM's current market lead in mobile over Intel - it is DESIGN."

    "Fabrication is not the reason for ARM's current market lead in mobile over Intel - it is DESIGN."

    Perhaps, but we're talking about a market that replaces handsets every two years. Like I said before when on phone is free on a two year contract and another one costs $200 consumers are going to give it the once over with a critical eye. Aside from iPhone I doubt consumers will take out their wallets for something slightly better.

    It's not the first shot you need to worry about from Intel. It's the ones after.

    Good luck being long on ARMH. Perhaps it will go to 50 or perhaps the "we're going to crush Intel" story has been a bit played out and you'll have a nice ride down to the single digits.

Add your comment.

DocumentId: 1870456, ~/Articles/ArticleHandler.aspx, 4/16/2014 1:46:42 AM

Report This Comment

Use this area to report a comment that you believe is in violation of the community guidelines. Our team will review the entry and take any appropriate action.

Sending report...


Advertisement